Table Of Contents

Previous topic

measures.glm

Next topic

measures.noiseperturbation

This content refers to the previous stable release of PyMVPA. Please visit www.pymvpa.org for the most recent version of PyMVPA and its documentation.

measures.irelief

Module: measures.irelief

Inheritance diagram for mvpa.measures.irelief:

FeaturewiseDatasetMeasure performing multivariate Iterative RELIEF (I-RELIEF) algorithm. See : Y. Sun, Iterative RELIEF for Feature Weighting: Algorithms, Theories, and Applications, IEEE Trans. on Pattern Analysis and Machine Intelligence (TPAMI), vol. 29, no. 6, pp. 1035-1051, June 2007.

Classes

IterativeRelief

class mvpa.measures.irelief.IterativeRelief(threshold=0.01, kernel_width=1.0, w_guess=None, **kwargs)

Bases: mvpa.measures.base.FeaturewiseDatasetMeasure

FeaturewiseDatasetMeasure that performs multivariate I-RELIEF algorithm. Batch version.

Batch I-RELIEF-2 feature weighting algorithm. Works for binary or multiclass class-labels. Batch version with complexity O(T*N^2*I), where T is the number of iterations, N the number of instances, I the number of features.

See: Y. Sun, Iterative RELIEF for Feature Weighting: Algorithms, Theories, and Applications, IEEE Trans. on Pattern Analysis and Machine Intelligence (TPAMI), vol. 29, no. 6, pp. 1035-1051, June 2007. http://plaza.ufl.edu/sunyijun/Paper/PAMI_1.pdf

Note that current implementation allows to use only exponential-like kernels. Support for linear kernel will be added later.

Note

Available state variables:

  • base_sensitivities: Stores basic sensitivities if the sensitivity relies on combining multiple ones
  • null_prob+: State variable
  • null_t: State variable
  • raw_results: Computed results before applying any transformation algorithm

(States enabled by default are listed with +)

See also

Please refer to the documentation of the base class for more information:

FeaturewiseDatasetMeasure

Constructor of the IRELIEF class.

Parameters:
  • enable_states (None or list of basestring) – Names of the state variables which should be enabled additionally to default ones
  • disable_states (None or list of basestring) – Names of the state variables which should be disabled
  • combiner (Functor) – The combiner is only applied if the computed featurewise dataset measure is more than one-dimensional. This is different from a transformer, which is always applied. By default, the sum of absolute values along the second axis is computed.
  • transformer (Functor) – This functor is called in __call__() to perform a final processing step on the to be returned dataset measure. If None, nothing is called
  • null_dist (instance of distribution estimator) – The estimated distribution is used to assign a probability for a certain value of the computed measure.
compute_M_H(label)

Compute hit/miss dictionaries.

For each instance compute the set of indices having the same class label and different class label.

Note that this computation is independent of the number of features.

XXX should it be some generic function since it doesn’t use self

k(distances)

Exponential kernel.

IterativeReliefOnline

class mvpa.measures.irelief.IterativeReliefOnline(a=10.0, permute=True, max_iter=3, **kwargs)

Bases: mvpa.measures.irelief.IterativeRelief

FeaturewiseDatasetMeasure that performs multivariate I-RELIEF algorithm. Online version.

This algorithm is exactly the one in the referenced paper (I-RELIEF-2 online), using weighted 1-norm and Exponential Kernel.

Note

Available state variables:

  • base_sensitivities: Stores basic sensitivities if the sensitivity relies on combining multiple ones
  • null_prob+: State variable
  • null_t: State variable
  • raw_results: Computed results before applying any transformation algorithm

(States enabled by default are listed with +)

See also

Please refer to the documentation of the base class for more information:

IterativeRelief

Constructor of the IRELIEF class.

Parameters:
  • enable_states (None or list of basestring) – Names of the state variables which should be enabled additionally to default ones
  • disable_states (None or list of basestring) – Names of the state variables which should be disabled
  • combiner (Functor) – The combiner is only applied if the computed featurewise dataset measure is more than one-dimensional. This is different from a transformer, which is always applied. By default, the sum of absolute values along the second axis is computed.
  • transformer (Functor) – This functor is called in __call__() to perform a final processing step on the to be returned dataset measure. If None, nothing is called
  • null_dist (instance of distribution estimator) – The estimated distribution is used to assign a probability for a certain value of the computed measure.

IterativeReliefOnline_Devel

class mvpa.measures.irelief.IterativeReliefOnline_Devel(a=5.0, permute=True, max_iter=3, **kwargs)

Bases: mvpa.measures.irelief.IterativeRelief_Devel

FeaturewiseDatasetMeasure that performs multivariate I-RELIEF algorithm. Online version.

UNDER DEVELOPMENT

Online version with complexity O(T*N*I), where N is the number of instances and I the number of features.

See: Y. Sun, Iterative RELIEF for Feature Weighting: Algorithms, Theories, and Applications, IEEE Trans. on Pattern Analysis and Machine Intelligence (TPAMI), vol. 29, no. 6, pp. 1035-1051, June 2007. http://plaza.ufl.edu/sunyijun/Paper/PAMI_1.pdf

Note that this implementation is not fully online, since hit and miss dictionaries (H,M) are computed once at the beginning using full access to all labels. This can be easily corrected to a full online implementation. But this is not mandatory now since the major goal of this current online implementation is reduction of computational complexity.

Note

Available state variables:

  • base_sensitivities: Stores basic sensitivities if the sensitivity relies on combining multiple ones
  • null_prob+: State variable
  • null_t: State variable
  • raw_results: Computed results before applying any transformation algorithm

(States enabled by default are listed with +)

See also

Please refer to the documentation of the base class for more information:

IterativeRelief_Devel

Constructor of the IRELIEF class.

Parameters:
  • enable_states (None or list of basestring) – Names of the state variables which should be enabled additionally to default ones
  • disable_states (None or list of basestring) – Names of the state variables which should be disabled
  • combiner (Functor) – The combiner is only applied if the computed featurewise dataset measure is more than one-dimensional. This is different from a transformer, which is always applied. By default, the sum of absolute values along the second axis is computed.
  • transformer (Functor) – This functor is called in __call__() to perform a final processing step on the to be returned dataset measure. If None, nothing is called
  • null_dist (instance of distribution estimator) – The estimated distribution is used to assign a probability for a certain value of the computed measure.

IterativeRelief_Devel

class mvpa.measures.irelief.IterativeRelief_Devel(threshold=0.01, kernel=None, kernel_width=1.0, w_guess=None, **kwargs)

Bases: mvpa.measures.base.FeaturewiseDatasetMeasure

FeaturewiseDatasetMeasure that performs multivariate I-RELIEF algorithm. Batch version allowing various kernels.

UNDER DEVELOPEMNT.

Batch I-RELIEF-2 feature weighting algorithm. Works for binary or multiclass class-labels. Batch version with complexity O(T*N^2*I), where T is the number of iterations, N the number of instances, I the number of features.

See: Y. Sun, Iterative RELIEF for Feature Weighting: Algorithms, Theories, and Applications, IEEE Trans. on Pattern Analysis and Machine Intelligence (TPAMI), vol. 29, no. 6, pp. 1035-1051, June 2007. http://plaza.ufl.edu/sunyijun/Paper/PAMI_1.pdf

Note that current implementation allows to use only exponential-like kernels. Support for linear kernel will be added later.

Note

Available state variables:

  • base_sensitivities: Stores basic sensitivities if the sensitivity relies on combining multiple ones
  • null_prob+: State variable
  • null_t: State variable
  • raw_results: Computed results before applying any transformation algorithm

(States enabled by default are listed with +)

See also

Please refer to the documentation of the base class for more information:

FeaturewiseDatasetMeasure

Constructor of the IRELIEF class.

Parameters:
  • enable_states (None or list of basestring) – Names of the state variables which should be enabled additionally to default ones
  • disable_states (None or list of basestring) – Names of the state variables which should be disabled
  • combiner (Functor) – The combiner is only applied if the computed featurewise dataset measure is more than one-dimensional. This is different from a transformer, which is always applied. By default, the sum of absolute values along the second axis is computed.
  • transformer (Functor) – This functor is called in __call__() to perform a final processing step on the to be returned dataset measure. If None, nothing is called
  • null_dist (instance of distribution estimator) – The estimated distribution is used to assign a probability for a certain value of the computed measure.
compute_M_H(label)

Compute hit/miss dictionaries.

For each instance compute the set of indices having the same class label and different class label.

Note that this computation is independent of the number of features.